Microsoft Increasing Use of In-House AI Chips as Alternative to Nvidia
Microsoft is shifting toward using more in-house developed artificial intelligence chips, aiming to reduce its reliance on Nvidia's graphics processing units. The company plans for the Maia family of AI accelerators—starting with the third-generation Maia 300 chip—to become a primary architecture by 2027. This move aligns with growing demand for advanced processors and diversification in silicon options across major tech firms. Microsoft has already introduced earlier versions of its Maia chips, including the Maia 100 (launched in 2023) and the Maia 200 (touted as delivering better performance per dollar). The upcoming Maia 300 will be deployed at a scale measured in gigawatts. Each data center rack is expected to house up to 96 of these chips, offering significantly higher AI token processing capabilities compared to GPUs. While Microsoft still partners with Nvidia for model training and other applications, the company's push toward self-developed silicon reflects broader industry trends where major cloud providers are seeking more control over their hardware infrastructure. Analysts suggest that Maia could represent around 10% of Microsoft’s AI chip spending by early 2027.
Tickers: $AMD $INTC $MSFT $NVDA
Topics: AI chips Maia chip architecture Microsoft Azure Nvidia alternatives
Part of the story Microsoft Launches New AI Decision Model (8 articles)